OpenCompass LLM - Reasoning (CompassBench 2601): leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 61.5 |
| 2 | GPT-5.2 (High) | 60.1 |
| 3 | Claude Opus 4.5 (Thinking) | 56.8 |
| 4 | Kimi K2.5 | 54.7 |
| 5 | Qwen 3 Max (2026-01-23) | 52.5 |
| 6 | O3 (2025-04-16) (High) | 51.5 |
| 7 | DeepSeek V3.2 | 51.2 |
| 8 | Claude Sonnet 4.5 (Thinking) | 50.7 |
| 9 | GLM-4.7 | 50.4 |
| 10 | GPT-OSS-120B (High) | 49.3 |
| 11 | Grok 4.1 | 49 |
| 12 | Tencent HY 2.0 Think | 48.3 |
| 13 | Ring-1T | 44.9 |
| 14 | Qwen 3 235B A22B 2507 (Thinking) | 44.4 |
| 15 | Kimi K2 (Thinking) | 44.1 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-llm-reasoning-compassbench-2601 · How It Works · Data refreshed daily, snapshot 2026-09-19.